[gst-devel] make clean didn't

Thomas Vander Stichele thomas at urgent.rug.ac.be
Wed Sep 25 05:29:01 CEST 2002


> After doing a "cvs update", i tried rebuilding.  gst-register was giving
> me crazy errors (below) so i did a make clean:
> 
> (process:15996): GLib-GObject-WARNING **: specified class size for type `GstMpeg2dec' is smaller than the parent type's `GstElement' class size
> 
> (process:15996): GStreamer-CRITICAL **: file gstelementfactory.c: line 148 (gst_element_factory_new): assertion `type' failed
> 
> However, the "make clean" didn't work completely because i'm configured
> for building fewer plugins:
> 
> joshua at emit:~/gst/gst-plugins$ find . -name .libs -print
> ./ext/aalib/.libs
> ./ext/alsa/.libs
> ./ext/dvdnav/.libs
> ./ext/http/.libs
> ./ext/mpeg2dec/.libs
> 
> i'm not building aalib, alsa, etc, so they aren't getting cleaned out
> but gst-register finds them because it searchs recursively.  This is
> bad.  Can we do something like this in the "make clean" rule?
> 
> rm -rf `find . -name .libs -print`

Hm, that's not the automake way to do it.  If you have those plug-ins 
built, it is because you did a previous configure and build with those 
plugins enabled.  It is your "duty" (in the automake way) to clean out 
those plug-ins if you want to reconfigure with different options.

So, you would be doing:

./configure
make
(realize we need fewer)
make clean
./configure --with-less-plugins
make

And this would leave you with the right structure.

So, I am against adding this rule, especially since the only thing it 
solves is a special case for someone developing in cvs mode only, and not 
generally useful.  In fact, it will probably cause problems anyway in 
different auto* uses.


Thomas

 -- 

The Dave/Dina Project : future TV today ! - http://davedina.apestaart.org/
<-*-                      -*->
Don't fear superboy is here
and it feels good to me
<-*- thomas at apestaart.org -*->
URGent, the best radio on the Internet - 24/7 ! - http://urgent.rug.ac.be/





More information about the gstreamer-devel mailing list